Inspiration
With the advent of the transformers the SOTA performance of computers in the NLP tasks such as summarisation has been quickly. This inspired me think to think that this accuracy can be used in improving the productivity of online classes.
What it does
Zoom Friend is an AI powered tool for automatically generating the transcript, summary and sample questions from any of your recorded class lectures.
How I built it
The backend for the application is made out of flask and the frontend is made up of regular HTML, CSS and JS. The NLP pipelines are served through Pytorch
Challenges I ran into
Gluing many pieces of varying technologies is always difficult and there were many versioning issues with the software packages that took a long time to figure out.
Accomplishments that I'm proud of
I am pound to be able to create an application which can improve the productivity of students in the COVID struck world.
What I learned
I learned to integrate ML pipelines into a full stack application.
What's next for Zoom Friend
There is a lot of scope of improvement. Firstly it should be hosted on a GPU powered server on one of the big cloud providers such as AWS or Azure


Log in or sign up for Devpost to join the conversation.